1a65d232b0
man-page: Move the PRETTY PRINTING header earlier
...
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2021-08-12 09:41:13 -03:00
2d35630fa5
pahole: Make pipe_seek() honour the 'fp' arg instead of hardcoding stdin
...
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2021-08-12 09:41:13 -03:00
9aa01472d9
pahole: Rename 'fp' to 'output' in prototype__stdio_fprintf_value()
...
As we'll also have another FILE pointer for input.
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2021-08-12 09:41:13 -03:00
472b940180
pahole: Use the supplied 'fp' argument in type__instance_read_once()
...
It was unconditionally reading from 'stdin', when a 'fp' is supplied.
Fix this as now we'll stop unconditionally reading from stdin for the
pretty printer.
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2021-08-12 09:41:13 -03:00
63992cb02a
core: Use namespace->name in class__clone()
...
Now that we stopped using string indexes, no need for that, just set
namespace->name with the new class name.
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2021-08-12 09:41:13 -03:00
ced4c34c37
core: Remove strings.c, unused
...
We were using this just for the ctf_encoder, that never really got
complete, so ditch it.
For BTF the strings table is done by libbpf, so we don't need it there
either.
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2021-08-12 09:41:11 -03:00
adbb66c295
ctf_loader: Use uint32_t instead of strings_t, that is going away
...
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2021-08-12 09:39:46 -03:00
ad707445b3
core: Remove base_type_name_to_size_table.sname, unused
...
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2021-08-12 09:39:46 -03:00
f8d571934b
pahole: Add missing bpf/btf.h include
...
We get it by accident, via pahole_strings.h, and that is going away, fix
it.
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2021-08-12 09:39:46 -03:00
deb6a4a492
libctf: Comment out unused CTF encoding functions
...
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2021-08-12 09:39:46 -03:00
82155bc92f
btf_loader: Add missing bpf/btf.h include
...
We get it by accident, via pahole_strings.h, and that is going away, fix
it.
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2021-08-12 09:39:46 -03:00
05f1f9dece
btf_encoder: Add missing bpf/btf.h include
...
We get it by accident, via pahole_strings.h, and that is going away, fix
it.
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2021-08-12 09:39:46 -03:00
f4a77d0390
pahole: Use conf_load.kabi_prefix
...
Should work just as before, i.e. we hook at wher we read strings from
DWARF.
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2021-08-12 09:39:46 -03:00
2b9bd83e63
dwarf_loader: Make attr_suffix() handle kabi_prefix
...
Since we're going to get rid of strings.c.
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2021-08-12 09:39:46 -03:00
daaafeb35f
dwarf_loader: Pass conf_load to functions calling attr_string()
...
As we'll implement that kabi_prefix thing there and without using global
variables.
This is because we're stopping usage of strings.c, where the kabi_prefix
feature was implemented.
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2021-08-12 09:39:46 -03:00
def39099c1
ctf_loader: No need for the 'strings' extern, not used
...
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2021-08-12 09:39:46 -03:00
2499920ac8
btf_loader: No need for the 'strings' extern, not used
...
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2021-08-12 09:39:46 -03:00
a388aaf489
dwarf_loader: Remove unused strings variable and debug_fmt_ops->{init,exit}()
...
No need to create that object anymore.
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2021-08-12 09:39:46 -03:00
3d3b7b3287
core: Remove unused debug_fmt_ops->dwarf__strings_ptr()
...
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2021-08-12 09:39:46 -03:00
05687c547e
core: Remove unused cu__string() method
...
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2021-08-12 09:39:46 -03:00
a201149e18
dwarf_loader: No need to strdup() what dwarf_formstring() returns
...
Conversation with Mark Wieelard, elfutils developer:
acme | ultimately dwarf_attr->valp for strings point to what? DIE memory that is always there?
acme | I'm working on pahole and need to keep a pointer to what it returns
acme | I was strdup()ing what dwarf_formstring() returns, tried removing the strdup() and instead point to what dwarf_formstring() returns and it worked, but I want to know for sure
mjw | ah, yeah
mjw | the memory/string returned by dwarf_formstring() is owned by the Dwarf. So long as the Dwarf is active (dwarf_end() hasn't been called and the underlying Elf is valid of course) you can use that string.
acme | cool!
acme | I'll use your explanation in the commit log message
acme | thanks!
acme | I'm also working on multithreading DWARF loading
mjw | in most cases it will point directly into the .debug_str section, but it can also be the .debug_line_str section or a string embedded in the .debug_info section, etc.
mjw | in all cases the Dwarf is responsible for keeping the memory life.
Before:
⬢[acme@toolbox pahole]$ rm -f vmlinux.btf ; perf stat -r5 pahole --btf_encode_detached vmlinux.btf vmlinux && perf stat -r5 btfdiff vmlinux vmlinux.btf
Performance counter stats for 'pahole --btf_encode_detached vmlinux.btf vmlinux' (5 runs):
7,802.91 msec task-clock:u # 0.989 CPUs utilized ( +- 0.60% )
0 context-switches:u # 0.000 /sec
0 cpu-migrations:u # 0.000 /sec
871,574 page-faults:u # 110.568 K/sec ( +- 0.00% )
29,924,977,089 cycles:u # 3.796 GHz ( +- 0.60% ) (83.32%)
455,561,473 stalled-cycles-frontend:u # 1.51% frontend cycles idle ( +- 5.55% ) (83.33%)
3,874,761,771 stalled-cycles-backend:u # 12.86% backend cycles idle ( +- 2.24% ) (83.34%)
74,812,680,221 instructions:u # 2.48 insn per cycle
# 0.05 stalled cycles per insn ( +- 0.02% ) (83.34%)
17,624,163,403 branches:u # 2.236 G/sec ( +- 0.03% ) (83.34%)
128,991,472 branch-misses:u # 0.73% of all branches ( +- 0.07% ) (83.33%)
7.8861 +- 0.0471 seconds time elapsed ( +- 0.60% )
Performance counter stats for 'btfdiff vmlinux vmlinux.btf' (5 runs):
6,323.23 msec task-clock:u # 1.000 CPUs utilized ( +- 0.97% )
0 context-switches:u # 0.000 /sec
0 cpu-migrations:u # 0.000 /sec
826,233 page-faults:u # 130.852 K/sec ( +- 0.00% )
23,719,098,640 cycles:u # 3.756 GHz ( +- 0.32% ) (83.35%)
286,636,981 stalled-cycles-frontend:u # 1.21% frontend cycles idle ( +- 2.52% ) (83.34%)
2,821,674,085 stalled-cycles-backend:u # 11.91% backend cycles idle ( +- 1.20% ) (83.28%)
64,095,069,092 instructions:u # 2.70 insn per cycle
# 0.04 stalled cycles per insn ( +- 0.03% ) (83.35%)
15,398,500,941 branches:u # 2.439 G/sec ( +- 0.02% ) (83.35%)
80,187,703 branch-misses:u # 0.52% of all branches ( +- 0.32% ) (83.34%)
6.3233 +- 0.0613 seconds time elapsed ( +- 0.97% )
⬢[acme@toolbox pahole]$
After:
static struct dwarf_off_ref attr_type(Dwarf_Die *die, uint32_t attr_name)
⬢[acme@toolbox pahole]$ rm -f vmlinux.btf ; perf stat -r5 pahole --btf_encode_detached vmlinux.btf vmlinux && perf stat -r5 btfdiff vmlinux vmlinux.btf
Performance counter stats for 'pahole --btf_encode_detached vmlinux.btf vmlinux' (5 runs):
7,008.59 msec task-clock:u # 0.977 CPUs utilized ( +- 1.03% )
0 context-switches:u # 0.000 /sec
0 cpu-migrations:u # 0.000 /sec
796,469 page-faults:u # 111.073 K/sec ( +- 0.00% )
28,167,752,342 cycles:u # 3.928 GHz ( +- 0.26% ) (83.32%)
377,704,478 stalled-cycles-frontend:u # 1.35% frontend cycles idle ( +- 0.96% ) (83.34%)
3,758,855,221 stalled-cycles-backend:u # 13.43% backend cycles idle ( +- 1.68% ) (83.34%)
72,453,367,989 instructions:u # 2.59 insn per cycle
# 0.05 stalled cycles per insn ( +- 0.03% ) (83.33%)
17,110,081,987 branches:u # 2.386 G/sec ( +- 0.02% ) (83.34%)
116,081,751 branch-misses:u # 0.68% of all branches ( +- 0.32% ) (83.33%)
7.1731 +- 0.0724 seconds time elapsed ( +- 1.01% )
Performance counter stats for 'btfdiff vmlinux vmlinux.btf' (5 runs):
5,768.59 msec task-clock:u # 1.014 CPUs utilized ( +- 0.45% )
0 context-switches:u # 0.000 /sec
0 cpu-migrations:u # 0.000 /sec
751,092 page-faults:u # 132.237 K/sec ( +- 0.00% )
21,623,439,905 cycles:u # 3.807 GHz ( +- 0.46% ) (83.34%)
221,665,165 stalled-cycles-frontend:u # 1.02% frontend cycles idle ( +- 1.55% ) (83.30%)
2,860,640,878 stalled-cycles-backend:u # 13.10% backend cycles idle ( +- 2.03% ) (83.32%)
61,757,937,981 instructions:u # 2.83 insn per cycle
# 0.04 stalled cycles per insn ( +- 0.01% ) (83.37%)
14,873,361,434 branches:u # 2.619 G/sec ( +- 0.02% ) (83.36%)
65,356,868 branch-misses:u # 0.44% of all branches ( +- 0.07% ) (83.35%)
5.6884 +- 0.0282 seconds time elapsed ( +- 0.50% )
⬢[acme@toolbox pahole]$
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2021-08-12 09:39:46 -03:00
6b7f1b72f9
core: No need for debug_fmt_ops->variable_name() anymore
...
Since we store a char pointer string in variable->name, we can use it
directly.
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2021-08-12 09:39:46 -03:00
9d0e3ab9a2
pahole: function__name() doesn't need a 'struct cu *' argument
...
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2021-08-12 09:39:46 -03:00
a7d789a4f8
core: Make variable->name a real string
...
For the threaded code we want to access strings in tags at the same time
that the string table may grow in another thread making the previous
pointer invalid, so, to avoid excessive locking, use plain strings.
The way the tools work will either consume the just produced CU straight
away or keep just one copy of each data structure when we keep all CUs
in memory, so lets try stopping using strings_t for strings.
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2021-08-12 09:39:46 -03:00
b5694280ec
core: Make label->name a real string
...
For the threaded code we want to access strings in tags at the same time
that the string table may grow in another thread making the previous
pointer invalid, so, to avoid excessive locking, use plain strings.
The way the tools work will either consume the just produced CU straight
away or keep just one copy of each data structure when we keep all CUs
in memory, so lets try stopping using strings_t for strings.
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2021-08-12 09:39:46 -03:00
e974d1b240
pahole: class_member_filter__new() doesn't need a 'struct cu *' argument
...
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2021-08-12 09:39:46 -03:00
0275e8d249
pahole: class_member_filter__parse() doesn't need a 'struct cu *' argument
...
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2021-08-12 09:39:46 -03:00
90183e8e4d
pahole: tag__real_sizeof() doesn't need a 'struct cu *' argument
...
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2021-08-12 09:39:46 -03:00
5cb9192738
pahole: Rename tag__fprintf_hexdump_value() to instance__fprintf_hexdump_value()
...
As it acts only on an instance, doesn't need neither a 'struct tag' nor
a 'struct cu'.
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2021-08-12 09:39:46 -03:00
75c769a900
pahole: enumerations__lookup_entry_from_value() doesn't need to return a CU anymore
...
As it will not be used in the caller.
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2021-08-12 09:39:46 -03:00
1edca26552
pahole: enumeration__lookup_entry_from_value() doesn't need a 'cu' argument
...
With the conversion of ->name members to plain char strings, no need
to use 'cu' to get the old string_t index and find the per-cu string
table.
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2021-08-12 09:39:46 -03:00
f8d98eff75
core: Ditch unused enumeration__prefix_len() method
...
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2021-08-12 09:39:46 -03:00
5cc365164a
core: Ditch unused enumeration__prefix() method
...
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2021-08-12 09:39:46 -03:00
e18c60d793
pahole: enumeration__lookup_value() doesn't need a 'cu' argument
...
With the conversion of ->name members to plain char strings, no need
to use 'cu' to get the old string_t index and find the per-cu string
table.
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2021-08-12 09:39:46 -03:00
4b877c8e67
pahole: enumeration__lookup_enumerator() doesn't need a 'cu' argument
...
With the conversion of ->name members to plain char strings, no need
to use 'cu' to get the old string_t index and find the per-cu string
table.
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2021-08-12 09:39:46 -03:00
3ff11828fe
core: enumeration__emit_definitions() doesn't need a 'cu' argument
...
With the conversion of ->name members to plain char strings, no need
to use 'cu' to get the old string_t index and find the per-cu string
table.
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2021-08-12 09:39:46 -03:00
0947d6e795
core: enumeration__fprintf() doesn't need a 'cu' argument
...
With the conversion of ->name members to plain char strings, no need
to use 'cu' to get the old string_t index and find the per-cu string
table.
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2021-08-12 09:39:46 -03:00
bb22f5bb0a
core: Make enumeration__max_entry_name_len() static
...
As it is not used outside where it is defined.
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2021-08-12 09:39:46 -03:00
dc83336171
core: enumeration__max_entry_name_len() doesn't need a 'cu' argument
...
With the conversion of ->name members to plain char strings, no need
to use 'cu' to get the old string_t index and find the per-cu string
table.
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2021-08-12 09:39:46 -03:00
ec1667c76d
core: Make enumeration__calc_prefix() static
...
As it is not used outside where it is defined.
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2021-08-12 09:39:46 -03:00
45ec63ed20
core: enumeration__calc_prefix doesn't need a 'cu' argument
...
With the conversion of ->name members to plain char strings, no need
to use 'cu' to get the old string_t index and find the per-cu string
table.
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2021-08-12 09:39:46 -03:00
2fae84e2f7
btf_encoder: btf_encoder__add_enum_type() doesn't need a 'cu' argument, ditch it
...
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2021-08-12 09:39:46 -03:00
96243fdd79
core: enumerator__name() doesn't need a 'cu' argument, ditch it
...
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2021-08-12 09:39:46 -03:00
ee5c12893b
core: Ditch dwarves__active_loader, unused
...
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2021-08-12 09:39:46 -03:00
713239bc00
core: Make enumerator->name a real string
...
For the threaded code we want to access strings in tags at the same time
that the string table may grow in another thread making the previous
pointer invalid, so, to avoid excessive locking, use plain strings.
The way the tools work will either consume the just produced CU straight
away or keep just one copy of each data structure when we keep all CUs
in memory, so lets try stopping using strings_t for strings.
For the enumerator->name case we get the bonus of removing the last user
of dwarves__active_loader in the btf_encoder class.
This covers unions, enums, structs and classes.
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2021-08-12 09:39:46 -03:00
7721cc17ac
emit: type__emit_fwd_decl() isn't used outside emit.c, make it static
...
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2021-08-12 09:39:46 -03:00
9aa5db7acd
emit: type__emit_fwd_decl() doesn't need a cu arg
...
Due to previous simplifications in the FOO__name() methods.
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2021-08-12 09:39:46 -03:00
c46f91ef6f
emit: type_emissions__find_definition() doesn't need a cu arg
...
Due to previous simplifications in the FOO__name() methods.
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2021-08-12 09:39:46 -03:00
c127d25daf
core: class__name() doesn't need a cu arg
...
Now that namespace->name is a real char string.
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2021-08-12 09:39:46 -03:00
00e8c5fe21
core: type__name() doesn't need a cu arg
...
Now that namespace->name is a real char string.
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2021-08-12 09:39:46 -03:00
Arnaldo Carvalho de Melo